OK it is not 100 per cent accurate but why are so many people who have no way of proving their innocence seemingly reluctant to take it.
The polygraph only measures certain physiological parameters, conductivity, blood pressure, etc., and thus are only really capable of measuring when a person becomes anxious. In use, subjects are told the polygraph is capable of detecting lies, which it simply isn't. However, imagine someone being accused of a heinous act by an intimidating interrogator. They're highly likely to be anxious in that situation, anyway, and can often give false positives when being interviewed. That's the case with innocent people.
In the case of criminals, however, do you not seriously believe they won't have found out how to fool the machine? It's extraordinarily simple to do. So we have a situation where the innocent might appear guilty through stress and fear while the guilty are capable of fooling the test.
The polygraph will not work effectively on psychopaths or sociopaths, either, or with many cases of Asperger syndrome or those with Autism.
So to summarise, it will show that innocent people telling the truth might be lying, but won't work at all with serious criminals. Finally, the scientific community in general has no confidence in it, while the
National Research Council has found no evidence of effectiveness. The utility among sex offenders is also poor with insufficient evidence to support accuracy or improved outcomes in this population.
In 2001 William Iacono, Professor of Psychology and Neuroscience concluded that
Although the CQT [Control Question Test] may be useful as an investigative aid and tool to induce confessions, it does not pass muster as a scientifically credible test. CQT theory is based on naive, implausible assumptions indicating (a) that it is biased against innocent individuals and (b) that it can be beaten simply by artificially augmenting responses to control questions.
Summarizing the consensus in psychological research, professor David W. Martin, PhD, from North Carolina State University, states that
people have tried to use the polygraph for measuring human emotions, but there is simply no royal road to (measuring) human emotions. Therefore, since one cannot reliably measure human emotions (especially when one has an interest in hiding his/her emotions), the idea of valid detection of truth or falsehood through measuring respiratory rate, blood volume, pulse rate and galvanic skin response is a mere pretense. Psychologists cannot ascertain what emotions one has, with or without the use of polygraph.
I could go on, Hugo, but there are numerous studies that have been done and are available on the internet which prove that there is no reliability with the Polygraph and - far from it 'not being 100% accurate' - your grandmother's left toe would have a better chance of ascertaining the credibility of a suspect.